Trajectories









Bookmarks

Lowest advantage
episodes
(unexpected failures):

trajectory 1, frame 145
trajectory 4, frame 418
trajectory 6, frame 318
trajectory 1, frame 473
trajectory 5, frame 445
trajectory 8, frame 455
trajectory 4, frame 123
trajectory 7, frame 246
trajectory 5, frame 301
trajectory 4, frame 492
trajectory 1, frame 225
trajectory 5, frame 77
trajectory 8, frame 119
trajectory 2, frame 99
trajectory 4, frame 9
trajectory 3, frame 230

Highest advantage
episodes
(unexpected successes):

trajectory 1, frame 481
trajectory 6, frame 366
trajectory 7, frame 302
trajectory 6, frame 145
trajectory 8, frame 464
trajectory 1, frame 1
trajectory 3, frame 265
trajectory 3, frame 151
trajectory 5, frame 353
trajectory 2, frame 146
trajectory 6, frame 261
trajectory 6, frame 503
trajectory 6, frame 410
trajectory 4, frame 284
trajectory 8, frame 410
trajectory 1, frame 289

Layers


Timeline

frame: 1 policy: next action: →
↙
←
↖
↓
no-op
↑
↘
→
↗
D
A
W
S
Q
E
fps
advantage
0.981
value function
8.51

Attribution

Observation Positive attribution Negative attribution

policy logits:

sums of policy logits:

Attribution legend

Click to expand feature
Hover to isolate

1
2
3
4
5
6
7
8
not
shown
residual
(everything
else)

Hotkeys

go backwards
go forwards
toggle play/pause

Select a feature

Feature visualization

zoom in zoom out
fewer patches more patches